Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Will x86 be dead in a decade?
- Thread starter Apple Fan 2008
- Start date
- Sort by reaction score
Will x86 be a visible part of consumer mass market electronics in a decade? I rather doubt it, but it’s not hard to imagine Intel turning things around. Again, not something I expect, but it wouldn’t be particularly remarkable if it happened.
Will there be no more x86 systems in operation or production?
Put it this way … never mind the massive financial institutions that still run on COBOL-based systems … if you do cutting-edge modern scientific computing, including machine learning and other forms of artificial intelligence, even if you’re using Python or the like … the libraries you’re calling are themselves using LAPACK, which is all Fortran. Fortran! Who would have imagined in the ’50s that Fortran would still be around today, let alone at the heart of actual thinking machines!?
And even supposing Intel and AMD both go belly-up for whatever inconceivable reason and nobody is making new physical x86 microprocessors, it’s a guarantee that there will be insanely mind-boggling amounts of critical legacy systems running on virtual x86 systems for far, far, far, far longer than a mere decade.
All this despite the fact that I’d be surprised if you could buy anything with an x86 chip in it at your local retail electronics store in a decade.
b&
Will there be no more x86 systems in operation or production?
Put it this way … never mind the massive financial institutions that still run on COBOL-based systems … if you do cutting-edge modern scientific computing, including machine learning and other forms of artificial intelligence, even if you’re using Python or the like … the libraries you’re calling are themselves using LAPACK, which is all Fortran. Fortran! Who would have imagined in the ’50s that Fortran would still be around today, let alone at the heart of actual thinking machines!?
And even supposing Intel and AMD both go belly-up for whatever inconceivable reason and nobody is making new physical x86 microprocessors, it’s a guarantee that there will be insanely mind-boggling amounts of critical legacy systems running on virtual x86 systems for far, far, far, far longer than a mere decade.
All this despite the fact that I’d be surprised if you could buy anything with an x86 chip in it at your local retail electronics store in a decade.
b&
There has only been one generational shift with Apple Silicon on the Mac, and that faced some delays due to Covid-related matters causing supply chain issues in SE Asia. Second, the M-series SoCs are by their very nature more complex than an A-series SoC, so it can take more time to design those. Either way, there have not been enough generational shifts with Apple Silicon on the Mac side to even attempt to determine what the release cycle will look like.
Intel foolishly though that creating an entirely new architecture for 64-bit was the way to go, forsaking any and all backwards compatibility. Meanwhile, AMD built their own set of 64-bit extensions to the x86 ISA, which is what Intel still licenses to this day. I wouldn't be surprised if this X86-S is an attempt to stick a knife into the back of AMD once and for all.
PowerPC’s not even dead! Even the 6502 is still in production.
RISC V+ is going to take over as dominate arch in 10 years. Mark my words.
I guess RISC-V will be next.
Gotta catch them all!
My guess is x86 is too heavily entrenched to be wiped out in ten years, but in twenty? If things keep becoming more mobile -- like for example if your desktop can be replaced by a lightweight, comfortable headset that keeps getting cheaper and better -- then yes, x86 would be in great danger. On the other hand, if desktops and laptops remain a mature market segment with performance iterations every year, I expect there'll remain a significant place for x86 as inertia keeps it from getting squeezed out.
Obviously not. The x86s proposal rips out a lot of that cruft and nobody seems to think it'll have much serious impact on anyone because nobody is using any of it anymore.
Marketing around incompatibility-FUD was great business for a while, but it's a boat anchor now. Intel has to choose if they want to sell an everything processor to the 3 banks and a DoD site still running code from the 70's, or if they want to sell processors into modern applications that value performance above all and don't mind updating their OS every decade or so. Carrying a bunch of transistors and logic forward that no modern OS or application needs is holding them back.
Apple is running x86 code close to native speeds on Arm under Rosetta2. I wouldn't be surprised if a future step is to start dropping hardware support for legacy instructions that account for disproportionate complexity in the decoders and schedulers.
I don’t think so.
My guess is Business and Enterprise alone will keep it going for a good bit longer than that.
My guess is Business and Enterprise alone will keep it going for a good bit longer than that.
I agree with you, but even if that was the only fallout, it wouldn't be so bad. As it turns out, Intel and x86 survived but the HP and Compaq C-suites bought into the plan as well and both PA-RISC and Alpha died as a result.
The idea started with HP, Intel coopted it and managed to kill off two competing architectures along the way. As expensive and embarrassing as that episode was, I'm not sure it turned out badly for Intel. They wound up with the IP for both Alpha and StrongARM, and didn't have to face PA-RISC in the server market any longer.
Last edited:
Power PC already stopped development in the middle of the 7th console gen. The only reason the Wii U still had it was because it had Wii hardware in it for backwards compatibility, and also because it's chips were in development before IBM pulled the plug on PPC. (Ironic since that PPC chip is one of the reasons the Wii U bombed hard and got jack for software support since everyone moved on to x86 with the PS4 and Xbox One)
Plus, PPC wasn't really used much outside of the Mac, some game consoles, and some servers. x86 eclipsed it in every way. Will ARM do the same to x86? Probably not, since neither architecture is better, they're both different with their own uses. It's not like x86 vs PPC where x86 was objectively superior.
Neither is the Z80, and it’s still in production as well. If those various unrelated chips aren’t dead after all these years, then 10 more years for x86 is very likely. Heck, maybe even 20.
I don't think that instinct was foolish. I think the two foolish things were, first, picking the wrong entirely new architecture and, second, not forsaking backwards compatibility.
VLIW is a dumb approach for general computing. I have no idea why they went that way. Maybe they felt that they needed to leapfrog the RISC world with something new. There's some really successful DSPs that use VLIW, but those are generally single threaded computation monsters with the ability to recompile code with each generation. The little details HP added just didn't overcome the challenges.
Second was insisting that they'd run x86 code natively. It was a massive undertaking that consumed way too much silicon just for compatibility's sake. And it was slow... They eventually smartened up and emulated x86 via software in future generations, but Itanic was already a sinking ship by then.
Add to that an irredeemable Intel culture and failure was assured.
Last edited:
That has less to do with the cost of x86 vs ARM and more todo with the cost of Apple hardware.
I guess RISC-V will be next.
I wouldn't bet against that happening, though it's unlikely in the short term. I'd almost guarantee that Apple has MacOs running on RISC-V, just like they had it on x86 years before switching to Intel. But I'd guess that's their emergency plan in case something goes badly wrong with ARM.
As for x86 becoming irrelevant in 10 years? It's been around since the 1970's, so I don't expect that it will be going anywhere soon.
That is patently absurd. IBM was not fully committed to PPC, and there were 64-bit PPC designs in the making that were promising in terms of P/W. Apple bought one of the companies that had such a design worked out and has used their engineers to advance 64-bit ARM (which has a lot of architectural similarity to PPC).
In fact, POWER10 is superior to x86 for data center implementations, capable of replacing scores of x86 boxes with a single unit for considerable energy savings. It could, in theory, easily supplant x86 in the high-end market with better P/W, which is important for servers and data centers, where legacy software is less of a factor. (POWER10 processor cores blow "hyperthreading" out of the water, capable of running 8 threads on each core.)
It's probably true Power10 is superior to anything Intel performance-wise, it's awesome hardware, but it has the same limitation ARM does, it doesn't run x86 software, so it runs in its own ecosystem and will never encroach on x86. I even have a Power9 i at work, great machine, runs different software than our Windows servers. DB performance is scary good though.
Data center and server setups are much less dependent on old x86 software. If you buy a POWER10 box, I am confident that IBM will see to it that it runs what you need.
Actually there's a lot of x86 code that runs against our database. We're just as dependent on x86 as Power9 i...